Merge branch 'master' into raghav/29359

Author: Raghavyadav17

.github/workflows/actions.yml

@@ -132,14 +132,7 @@ jobs:
           pip install -r requirements-openvino.txt --progress-bar off --upgrade
           pip uninstall -y keras keras-nightly
           pip install -e "." --progress-bar off --upgrade
-      - name: Lint
-        run: bash shell/lint.sh
-      - name: Check for API changes
-        run: |
-          bash shell/api_gen.sh
-          git status
-          clean=$(git status | grep "nothing to commit")
-          if [ -z "$clean" ]; then
-            echo "Please run shell/api_gen.sh to generate API."
-            exit 1
-          fi
+      - name: Install pre-commit
+        run: pip install pre-commit && pre-commit install
+      - name: Run pre-commit
+        run: pre-commit run --all-files --hook-stage manual

.pre-commit-config.yaml

@@ -0,0 +1,31 @@
+repos:
+  - repo: local
+    hooks:
+      - id: api-gen
+        name: api_gen
+        entry: |
+          bash shell/api_gen.sh
+          git status
+          clean=$(git status | grep "nothing to commit")
+          if [ -z "$clean" ]; then
+            echo "Please run shell/api_gen.sh to generate API."
+            exit 1
+          fi
+        language: system
+        stages: [pre-commit, manual]
+        require_serial: true
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.9.2
+    hooks:
+      - id: ruff
+        args: [--config, pyproject.toml, --fix, .]
+        stages: [pre-commit]
+      - id: ruff-format
+        args: [--config, pyproject.toml, .]
+        stages: [pre-commit]
+      - id: ruff
+        args: [--config, pyproject.toml, .]
+        stages: [manual]
+      - id: ruff-format
+        args: ["--check", --config, pyproject.toml, .]
+        stages: [manual]

CONTRIBUTING.md

@@ -107,22 +107,35 @@ You can also add GPU support to your environment, see the
 [Adding GPU support](https://github.com/keras-team/keras/blob/master/README.md#adding-gpu-support)
 section of the README.
 
-## Code style
+## Generating public API and formatting the code
 
-Keras uses [Ruff](https://docs.astral.sh/ruff/) to format the code. Please refer to
-[requirements-common.txt](https://github.com/keras-team/keras/blob/master/requirements-common.txt)
-for the required versions. Run the following command **at the root directory of
-the repo** to format your code.
+For the first time you are setting up the repo, please run `pre-commit install`.
+Note that this needs to be done only once at the beginning.
+
+Now, whenever you run `git commit -m "<message>"`, three things are
+automatically done:
+
+- Public API generation
+- Code formatting
+- Code linting
+
+If there's any error, the commit will not go through. Please fix the error (
+most of the times, the error is fixed automatically by the formatter/linter) and
+re-run the following:
 
 ```
-sh shell/format.sh
+git add .
+git commit -m "<message>" # This will not get logged as a duplicate commit.
 ```
 
-It will also display the errors that cannot be resolved by autoformatting. You
-need to follow the output of the command to resolve them manually.
+In case you want to run the above manually on all files, you can do the
+following:
+
+```
+pre-commit run --all-files
+```
 
-If you do not want to auto format the code but only show the lint errors, you
-can run `sh shell/lint.sh` **at the root directory of the repo**.
+KerasHub uses [Ruff](https://docs.astral.sh/ruff/) to format the code.
 
 ### Docstrings
 

keras/api/_tf_keras/keras/losses/__init__.py

@@ -14,6 +14,7 @@
 from keras.src.losses.losses import BinaryFocalCrossentropy
 from keras.src.losses.losses import CategoricalCrossentropy
 from keras.src.losses.losses import CategoricalFocalCrossentropy
+from keras.src.losses.losses import CategoricalGeneralizedCrossEntropy
 from keras.src.losses.losses import CategoricalHinge
 from keras.src.losses.losses import Circle
 from keras.src.losses.losses import CosineSimilarity
@@ -34,6 +35,7 @@
 from keras.src.losses.losses import binary_focal_crossentropy
 from keras.src.losses.losses import categorical_crossentropy
 from keras.src.losses.losses import categorical_focal_crossentropy
+from keras.src.losses.losses import categorical_generalized_cross_entropy
 from keras.src.losses.losses import categorical_hinge
 from keras.src.losses.losses import circle
 from keras.src.losses.losses import cosine_similarity

keras/api/_tf_keras/keras/optimizers/__init__.py

@@ -19,6 +19,7 @@
 from keras.src.optimizers.lamb import Lamb
 from keras.src.optimizers.lion import Lion
 from keras.src.optimizers.loss_scale_optimizer import LossScaleOptimizer
+from keras.src.optimizers.muon import Muon
 from keras.src.optimizers.nadam import Nadam
 from keras.src.optimizers.optimizer import Optimizer
 from keras.src.optimizers.rmsprop import RMSprop

keras/api/losses/__init__.py

@@ -13,6 +13,7 @@
 from keras.src.losses.losses import BinaryFocalCrossentropy
 from keras.src.losses.losses import CategoricalCrossentropy
 from keras.src.losses.losses import CategoricalFocalCrossentropy
+from keras.src.losses.losses import CategoricalGeneralizedCrossEntropy
 from keras.src.losses.losses import CategoricalHinge
 from keras.src.losses.losses import Circle
 from keras.src.losses.losses import CosineSimilarity
@@ -33,6 +34,7 @@
 from keras.src.losses.losses import binary_focal_crossentropy
 from keras.src.losses.losses import categorical_crossentropy
 from keras.src.losses.losses import categorical_focal_crossentropy
+from keras.src.losses.losses import categorical_generalized_cross_entropy
 from keras.src.losses.losses import categorical_hinge
 from keras.src.losses.losses import circle
 from keras.src.losses.losses import cosine_similarity

keras/api/optimizers/__init__.py

@@ -19,6 +19,7 @@
 from keras.src.optimizers.lamb import Lamb
 from keras.src.optimizers.lion import Lion
 from keras.src.optimizers.loss_scale_optimizer import LossScaleOptimizer
+from keras.src.optimizers.muon import Muon
 from keras.src.optimizers.nadam import Nadam
 from keras.src.optimizers.optimizer import Optimizer
 from keras.src.optimizers.rmsprop import RMSprop

keras/src/backend/common/variables.py

@@ -96,6 +96,7 @@ def __init__(
         trainable=True,
         autocast=True,
         aggregation="none",
+        synchronization="auto",
         name=None,
     ):
         name = name or auto_name(self.__class__.__name__)
@@ -113,13 +114,28 @@ def __init__(
             "only_first_replica",
         ):
             raise ValueError(
-                "Invalid valid for argument `aggregation`. Expected "
+                "Invalid value for argument `aggregation`. Expected "
                 "one of `None`, `'none'`, `'mean'`, `'sum'`, "
                 "`'only_first_replica'`. "
                 f"Received: aggregation={aggregation}"
             )
         if aggregation is None:
             aggregation = "none"
+        if synchronization not in (
+            None,
+            "none",
+            "on_read",
+            "on_write",
+            "auto",
+        ):
+            raise ValueError(
+                "Invalid value for argument `synchronization`. Expected "
+                "one of `None`, `'none'`, `'on_read'`, `'on_write'`, "
+                "`'auto'`. "
+                f"Received: synchronization={synchronization}"
+            )
+        if synchronization is None:
+            synchronization = "none"
         self._name = name
         parent_path = current_path()
         if parent_path:
@@ -133,6 +149,7 @@ def __init__(
         self._trainable = bool(trainable)
         self._autocast = bool(autocast)
         self._aggregation = aggregation
+        self._synchronization = synchronization
         # `self._overwrite_with_gradient` is an internal property to determine
         # whether this variable should be overwritten by the computed gradient.
         # Ref: https://github.com/google/flax/blob/main/flax/linen/fp8_ops.py
@@ -227,6 +244,11 @@ def aggregation(self):
         """The strategy for aggregating this variable."""
         return self._aggregation
 
+    @property
+    def synchronization(self):
+        """The strategy for synchronizing this variable."""
+        return self._synchronization
+
     @property
     def value(self):
         """The current value of the variable (numpy array or backend tensor)."""

keras/src/backend/openvino/excluded_concrete_tests.txt

@@ -99,7 +99,9 @@ NumpyOneInputOpsCorrectnessTest::test_hstack
 NumpyOneInputOpsCorrectnessTest::test_imag
 NumpyOneInputOpsCorrectnessTest::test_isfinite
 NumpyOneInputOpsCorrectnessTest::test_isinf
-NumpyOneInputOpsCorrectnessTest::test_log
+NumpyOneInputOpsCorrectnessTest::test_log1p
+NumpyOneInputOpsCorrectnessTest::test_log2
+NumpyOneInputOpsCorrectnessTest::test_logaddexp
 NumpyOneInputOpsCorrectnessTest::test_max
 NumpyOneInputOpsCorrectnessTest::test_mean
 NumpyOneInputOpsCorrectnessTest::test_median

keras/src/backend/openvino/numpy.py

@@ -841,11 +841,24 @@ def linspace(
 
 def log(x):
     x = get_ov_output(x)
+    x_type = x.get_element_type()
+    if x_type.is_integral():
+        x_type = OPENVINO_DTYPES[config.floatx()]
+        x = ov_opset.convert(x, x_type)
     return OpenVINOKerasTensor(ov_opset.log(x).output(0))
 
 
 def log10(x):
-    raise NotImplementedError("`log10` is not supported with openvino backend")
+    x = get_ov_output(x)
+    x_type = x.get_element_type()
+    if x_type.is_integral():
+        x_type = OPENVINO_DTYPES[config.floatx()]
+        x = ov_opset.convert(x, x_type)
+    log_x = ov_opset.log(x).output(0)
+    const_10 = ov_opset.constant(10, x_type).output(0)
+    log_10 = ov_opset.log(const_10).output(0)
+    result = ov_opset.divide(log_x, log_10).output(0)
+    return OpenVINOKerasTensor(result)
 
 
 def log1p(x):